Artificial intelligence curriculum for primary and secondary schools
Ju and Pi found themselves today in a vast and somewhat mysterious universe of words. Each word has its own place here, and is surrounded by many other words that are somehow related. And, as it turns out, even ordinary word-clubs can get the circuits pretty tangled! Their mission? To help Cat Cat, who is looking sad, and find out what can cheer her up. And they may also discover that the most important thing is not just choosing the right words, but how and to whom we dedicate them.
The lesson begins with a story in which the robots Ju and Pi find themselves in a "universe of words" - a vast space where every word has its place and is related to other words. Their task is to find out why their cat, Puss, is sad, and they try to find the answer using a language model. The teacher leads a discussion about how language models work - that they learn from large amounts of text and generate answers based on the probability of words occurring. Children try a simple game of adding words to a story, where they recognise that different choices can lead to different meanings. In a hands-on activity, they create their own "word universe" where they connect pictures and words according to their proximity in meaning, learning the principles of how language models organize information. If there is time left over, the lesson continues by experimenting with a chatbot, where students check how the models generate responses and discuss why they may sometimes produce nonsensical or incorrect sentences. Finally, children reflect on their knowledge of language models, their possibilities and limitations, and realize the importance of critical thinking when working with generative AI.
45-90 minutes, grades 3-5 Elementary school
Teacher: projection equipment, presentation to show, account in text generating application
Students: writing utensils, glue, quarter, printed worksheets
Language model
A language model is a program that has learned from many texts how people communicate. But sometimes it generates the wrong things.
Based on their understanding of how language models work, they critically assess the outputs of generative intelligence.
They will describe in their own words how the language model works and explain why it can generate incorrect things.
Informatics:
Data, Information, Modeling:
I-5-1-03 reads information from the given model
Benefits and development - understands the importance of digital technologies for human society, learns about new technologies, critically evaluates their benefits and reflects on the risks of their use.
Understanding: students will describe the principle of a language model and explain why it can generate incorrect things.
Creating: they will create their simplified model.
4-B-I Common sense reasoning